Inspiration:
A few months ago, 5 Indian migrant workers were kidnapped in Niger. The GOI struggled to trace them back because of no centralized records existed about their work or location. This is not an isolated case — millions of migrant workers face scams, unpaid wages, and unsafe working conditions due to missing records and lack of awareness of their rights.
This inspired me to build Pravas – an AI-powered agent that ensures migrant workers can Migrate Freely, with dignity and safety.
What it does:
Creates a digital identity & work record for every migrant. Lets workers upload job contracts as photos/PDFs → AI reviews them in seconds, highlighting anomalies or unfair terms. Explains worker rights in local languages clearly and simply. Detects unpaid wages, excessive work hours, or fraud automatically. Provides a one-tap SOS button to alert govt/NGOs with worker details, location, and employer info.
How we built it:
Frontend: Mobile-first app (React Native/Flutter). Backend: Node.js + Express with PostgreSQL for secure storage. AI Agent Modules: OCR + NLP for contract scanning and anomaly detection. Anomaly Models to check wage/work mismatches. Multilingual NLU to guide workers about rights in local dialects. Emergency SOS: Integrated location APIs + SMS/WhatsApp for alerts.
Challenges we ran into:
Designing for semi-literate users with minimal smartphone access. Ensuring privacy & trust while recording sensitive worker data. Building multilingual support across different accents/dialects. Simulating govt/NGO dashboards without actual integration. Avoiding false positives in anomaly detection.
Accomplishments that we're proud of:
Building an AI contract reviewer that flags anomalies within seconds. Creating a voice-first prototype to support non-literate workers. Designing a system that’s both scalable for government adoption and usable by grassroots NGOs. Framing AI as an active agent for worker safety, not just a passive tool.
What we learned:
AI agents can actively safeguard vulnerable groups, beyond just analytics. User empathy is key — designing for workers who may not read/write well. Balancing data security with usability is crucial for trust. Real-world deployment needs offline-first solutions (e.g., SMS/IVR).
What’s next for Pravas:
Partner with state labor departments & NGOs for pilot testing. Expand multilingual support for 10+ Indian languages. Add blockchain-backed digital work history for tamper-proof records. Explore voice-bot integration on WhatsApp/IVR for broader accessibility. Collaborate with policy-makers to integrate Pravas with e-Shram and other government schemes.
